Let’s get acquainted first, since I’m new in these parts and consider it necessary to introduce myself. For those who are not interested, skip straight to the next paragraph. Although I have known Stopgame for a relatively long time, this is my first time using the site. Since I passed the test, and a terrible storm is raging outside cordyceps coronavirus, I thought why not start a blog? Game journalism was my childhood dream, my specialty at the university overlaps with this, so why not try? Moreover, some of my work at university is related to video games, so I thought why not adapt it for the blog, although this article is not related to that. And yes, I’m one of those people who watched stopgame videos not because I’m too lazy to write coursework, but because I have to write coursework.

The words “machine learning” and “neural networks” can already be confidently called one of the main topics in science in recent years. Moreover, these are the very “cool” topics that the public willingly comes to. Just look at those funny videos where neural networks replace people’s faces.

The prospects for the https://nonukcasinosites.co.uk/welsh-not-on-gamstop/ development of this industry will be of particular interest to players. Game engine developers teach neurons to play games, make levels, and even create stories. If all goes well, the gaming industry of a bright future will greatly reduce the amount of monotonous work. AI will transfer the “Press x to win” principle into game development. And today I want to talk about one of these steps into the future.

So what’s up?

I am writing this material entirely based on the video from the Two Minute Papers channel and attaching it to the article, and for those who are not fluent in English and/or do not want to watch the video, I will briefly summarize its content.

The point is that the artificial intelligence was first fed a video where various liquids and gases were transformed according to the laws of physics, and then the viewing was stopped and the AI ​​was told to continue. Then they compared what the network did based on the video and what should have happened according to mathematical calculations. Surprisingly, the machine simulates the movement of various fluids quite plausibly (mathematical simulation on the left, AI prediction on the right)

Moreover, the network also copes well with complex figures and their interactions.

But why then should we use a neural network for this, if existing mathematical models cope with this perfectly, and machine learning is a long and complex process? The point is that as soon as we have trained the machine, the labor costs for miscalculation become much less than with formulas. Returning to the wonderful world of the future, the development of this technology will allow us to simulate various complex liquids in real time, taking into account their density, viscosity, the surface on which they are located, and at the same time the processor will not create a local Chernobyl on the motherboard. The age of a two-dimensional fire sprite instead of a real simulation will become the same archaism. Imagine how water, which is much more common, can change! Starting from realistic calculation of waves, ending with believable flow of water from different textures of clothing, depending on the material.

Conclusions

In this short article I wanted to share another step that neural networks have taken to help in game development. Artificial intelligence has penetrated our lives, but not in the way we expected. Instead of robots that feel emotions and can think independently, we get a universal assistant that deals with routine. And I think it’s wonderful.